Raspberry Pi Holdings plc (RPI) — Investment Research Note
Executive summary
Raspberry Pi designs and sells low-cost, high-performance single-board computers, compute modules and microcontrollers to industrial OEMs (~70% of demand), enthusiasts/education and via a licensee channel, supported by in-house semiconductor IP. The trajectory across the period is one of post-pandemic normalisation: 41% revenue growth in FY2023, a 2% revenue dip and 15% Adj EBITDA decline in FY2024 from channel destocking, then a strong FY2025 recovery (Adj EBITDA ≥$45m, +20%) followed by a powerful H1 2026 beat (Adj EBITDA ≥$38m vs $42m FY consensus) on cheap pre-bought DRAM. The single most important valuation point today is that the shares have ~4× in six months on this DRAM-arbitrage windfall, leaving the stock at roughly 50× sustainable EBITDA — heroic assumptions on AI-adjacent semis multiples are now embedded.
Fair value estimate
- Methodology: Forward EV/EBITDA (cross-checked against EV/Sales and P/E), centred on a "post-windfall" sustainable EBITDA estimate that strips out the temporary low-cost-DRAM inventory benefit which management itself flags as moderating in H2 2026.
- Key assumptions:
- Sustainable run-rate Adj EBITDA $50–65m (above FY2025's $45m, reflecting volume growth and OEM design-wins, but below the H1 2026 windfall-inflated rate). At ~£0.78/$ that's £39–51m.
- Multiple range 22–32× EBITDA (mid-cap UK tech/semis with growth but cyclical end-markets and ~24% gross margins).
- Net cash ~£20–25m.
- Fair value range: ~520p – 920p per share, midpoint ~720p, implying market cap range ~£1,000m – £1,780m, midpoint ~£1,400m.
- Latest disclosed market cap £2,034m versus midpoint £1,400m → ~31% downside to mid; range −50% to −13% vs the current 1,051p.
- The bull-case scenario (sustainable $80m EBITDA × 35×) gets to ~£2.2bn or ~1,135p — i.e. you have to assume the windfall persists structurally and the market awards a high-growth semis multiple to just justify the current price.
Sector context
- Sector: Technology (ICB Tech / Tech Hardware & Equipment). Confirmed — full-stack semiconductor IP + electronic product design + software.
- Quality/growth/leverage profile is above typical UK industrial-tech peers on balance sheet (net cash, fortress liquidity, $80m undrawn RCF 2025-04 FY2024 results) and on growth momentum, but below pure-play AI infrastructure semis on gross margin (24% vs 50–70% for fabless designers) and on AI revenue exposure.
- Comparable listed peers: limited direct comps; loosely Arm Holdings (semi IP, ecosystem partner and shareholder), Premier Farnell parent Avnet/Element14 (distributor), and on the embedded compute side Advantech / Kontron AG. None map perfectly.
Investment thesis (3 bullets)
- Inventory positioning is converting an industry headwind into a sustained earnings beat. Memory vendors are diverting LPDDR4 capacity to AI data centres, but RPI's long-standing buffer policy plus opportunistic 2025 purchases means H1 2026 Adj EBITDA already approximates the full-year consensus, with FY2026 EBITDA flagged "significantly ahead of current market expectations" 2026-06-05 trading update; 2026-01-13 year-end update.
- OEM/Industrial channel strategy is broadening the customer base beyond enthusiasts. Industrial/embedded is ~70% of demand; multi-year SECO strategic partnership signed, Compute Module 5 design wins, growing Approved Reseller network (+13 in 2024), and Connect for Organisations starting to monetise software/services 2025-04-02 FY2024 results; 2025-01-29 trading update.
- Balance-sheet fortress + founder-led, vertically integrated R&D. Net cash, $80m undrawn RCF extended to 2029, 67% of R&D capitalised on a rapidly expanding pipeline (RP2350 launched, RP1 deployed, next-gen flagship in development), Arm a strategic shareholder (and £50m participant in April 2026 placing) 2025-04 FY2024; 2026-04-23 secondary placing.
Key risks (3 bullets)
- Memory cost normalisation will erase the windfall. Management is explicit that H2 2026 unit economics will moderate as cheap DRAM depletes and the company has to absorb higher input costs; DRAM availability at high densities is constrained by AI capex 2026-06-05 trading update; 2026-01-13 year-end update.
- Single-supplier concentration on Broadcom SOC and Sony South Wales manufacturing. Any disruption would directly hit production and revenue; some board-level commentary acknowledges this is mitigated only by inventory buffers and contingency planning 2024-09 H1 2024 interim risks.
- Customer concentration in the channel. One major distributor was 27% of FY2024 revenue and the contract manufacturer 14%; channel destock in 2024 illustrated how quickly volumes can swing 2025-04 FY2024 final results, Revenue note.
Operating leverage
RPI's leverage profile is mid-tier hardware — better than a pure box-shifter but capped by ~24% gross margins. The fixed-cost base is modest: in FY2024, adjusted R&D was just $8.7m and adjusted admin $17.3m on $259m revenue, i.e. ~10% of revenue 2025-04 FY2024 results. A 10–20% revenue beat at the prevailing ~24% incremental gross margin would add $6–12m to gross profit; with limited opex creep that drops straight to EBITDA — roughly +15–30% on a ~$45m EBITDA base. The H1 2026 beat illustrates a sharper version of this leverage, but it's working through gross-margin uplift (favourable DRAM inventory), not through fixed-cost dilution — and that is by management's own admission temporary. There is no SaaS-style or platform-economics inflection here: Connect for Organisations has only "first paying subscribers in Q1 2025" and is immaterial today. The leverage is real but the multiple-of-profit kind the investor seeks would require either a step-change in gross margin (semiconductor IP/royalty mix shift) or scale to a much larger fixed cost base.
Value-trap signals
None identified. Growth has resumed (FY2025 EBITDA +20%, H1 2026 materially ahead), no dividend cuts (no dividend policy yet), no guidance misses, balance sheet is fortress, founder is engaged. The opposite signal — priced for perfection — is the live concern, not value-trap risk.
Earnings vs. expectations
Limited public history (IPO June 2024). The pattern observed:
- H1 2024 (Sept 2024 interim): profit "ahead of internal expectations", though full-year unchanged → modest beat 2024-09 interim.
- FY2024 (April 2025): Adj EBITDA $37.2m, "in line with guidance" → met 2025-04 FY2024 results.
- FY2025 (Jan 2026 trading update): Adj EBITDA ≥$45m vs consensus $40.9m → beat by ~10% 2026-01-13.
- H1 2026 (June 2026 trading update): Adj EBITDA ≥$38m, full-year flagged "significantly ahead" of $42m consensus → material beat / upgrade 2026-06-05. Pattern: consistent beats since IPO, with the most recent two being substantial. However the recent beats are heavily driven by the temporary DRAM inventory tailwind, which limits the read-across to underlying earnings power.
Conviction
3 — moderate.
Anchoring my view (supportive of conviction): clean post-IPO disclosure; founder-led management with credible commentary; explicit guidance on what is temporary (DRAM windfall) vs. sustainable.
Limiting conviction: (a) only two years of public reporting, so a multi-cycle track record is absent; (b) sustainable EBITDA is genuinely hard to pin down — the windfall portion of H1 2026 could be anywhere from $5m to $20m, materially shifting the multiple; (c) the appropriate exit multiple is contestable — peers range from 15× (commodity hardware) to 35×+ (AI-adjacent semis), and the market is currently anchoring to the higher band.